Compare the following talkback reports:
http://talkback-public.mozilla.org/reports/firefox/FF2a1/smart-analys...
http://talkback-public.mozilla.org/reports/firefox/FF2a2/smart-analys...

FF2a1 reports 88 unique users, and FF2a2 reports only 22 unique users.
 Are we just more stable now, or do we have fewer users?  If the
former than great, but if the latter, then perhaps it is time to offer
updates to Bon Echo alpha users to get them onto the newer alphas.

It would be nice to see how many talkback users are on FF2a3, but
unfortunately that data does not appear to be available.

-Darin

On 6/8/06, Darin Fisher <dar...@gmail.com> wrote:

Yes, I think it is.  Even now, a while after the a3 release, I think
it would be _immensely_ valuable to get those a1/a2 testers onto a3
via the update channel.  We want them testing what's current, for so
many reasons.

Mike

We should check to see if all the reports are hooked up to the right builds.

There was a big spike in "2.0 branch" nt.ddl crashes on May 31.  That
might have corresponded to more usage of alpha 3 when folks returned
from the Memorial day holiday in the US.

http://talkback-public.mozilla.org/reports/firefox/FF2x/FF2x-topcrash...

Releasing the alpha on a friday before a long weekend also might have
kept the visiblity of the release lower than normal.

The talkback numbers do indeed look very low and too low to really
understand what problems might exist.  Does any one know what download,
update, and usage numbers look like?

chris h.

Filed bug 340862 (https://bugzilla.mozilla.org/show_bug.cgi?id=340862)
to track this request.

--
/ mike beltzner / phenomenologist / mozilla corporation /

The nightlies seem to have a larger sampling.
http://talkback-public.mozilla.org/reports/firefox/FF2x/smart-analysi...
 Total unique users:     485
http://talkback-public.mozilla.org/reports/firefox/FFTrunk/smart-anal...
 Total unique users:     516

On 6/8/06, Darin Fisher <dar...@gmail.com> wrote:

This is probably due to turning back on Mork autocomplete. See bug 339861.

This is a huge problem, and I've been so far unable to reproduce it. It
seems that basically nobody has done any testing. Hopefully this post
will cause somebody that does see this problem to help out.

Since this just re-enabled old working code, I suspect that one of the
changes checked in on or after Storage autocomplete was first enabled is
causing the problem. We would not have noticed the problems because it
was disabled.

There are two options:

Option 1: Revert back to Storage form autocomplete. This should
immediately fix the problem, but will ensure that Mork autocomplete will
stay permanently broken because nobody is testing it. The storage
autocomplete is also not as well tested, and needs some work before
release for some things like recovering in case of a corrupted database
(currently it will just be permanently broken).

Option 2: Find the problem and fix the Mork version. This has the
disadvantage of taking longer and pissing people off.

I'm OK with either of these options. However, if we do (1), I feel that
we should delete Mork autocomplete from toolkit, which I imagine people
will be opposed to.

Brett